Teaching and Learning Research Report Series

The Academic Quality

of Prospective Teachers:

The Impact of
Admissions and
Licensure Testing

CONTENTS

This report was written by

Drew H. Gitomer and
Andrew S. Latham of
Educational Testing Service,
and by Robert Ziomek of
ACT, Inc. All responsibility
for the content of this report
rests solely with the authors

Additional copies can be downloaded

from the Praxis Web site at
www.ets.org/praxis/ or can be
ordered for $15.00 from:

The Teaching and Learning Division

Research and Data Analysis Group
Mail Stop 15-D
Educational Testing Service
Rosedale Road
Princeton, NJ 08541
Email: tandlresearch@ets.org

Copyright 1999 by Educational

Testing Service. All rights reserved.
EDUCATIONAL TESTING SERVICE,
ETS, and the ETS logo are registered
trademarks of Educational Testing
Service. The modernized ETS logo,
the TEACHING AND LEARNING
logo, and THE PRAXIS SERIES are
trademarks of Educational Testing
Service. College Board and SAT are
registered trademarks of the College
Entrance Examination Board. All
other registered trademarks,
trademarks, and service marks are
the property of their respective
owners. Educational Testing Service
is an Equal Opportunity/Affirmative
Action Employer.

ETS RR-03-35

Acknowledgements ......................................................................... 2
Abstract ........................................................................................... 3
The Context:Teacher Academic Ability, Supply, and Diversity ........ 4
The Impact of Teacher Testing .................................................... 7
Reforms Within Colleges of Education ....................................... 7
The Study ........................................................................................ 9
Key Issues ................................................................................... 9
Study Overview ........................................................................ 11
Limitations and Caveats ............................................................ 12
Study Findings ............................................................................... 14
SAT/ACT Baseline Data ........................................................... 14
Results For Those Entering Teacher Education Programs ......... 15
Results For Those Who Seek Licensure .................................... 19
Teacher Education Institutions ................................................. 24
Results by Licensure Area ......................................................... 26
Raising and Lowering Standards ............................................... 31
Conclusions ................................................................................... 38
References ..................................................................................... 41
Appendix A ................................................................................... 44

ACKNOWLEDGEMENTS
This study could not have been done without the generous help of many
individuals and organizations. The American Association for Colleges of
Teacher Education encouraged us to take on these challenging issues. We are
especially grateful to Mary Dilworth, David Imig, and AACTEs Research and
Information Committee for their encouragement and insight during the past
year.
Our home organizations have supported this research both financially and
intellectually. The ETS authors are especially grateful to Mari Pearlman,
Barbara Vilkomerson, Marisa Farnum, Sharon Robinson, and Nancy Cole for
creating the opportunity to undertake this work and for their wise advice
throughout. Likewise, we want to thank Nancy Petersen of ACT for
encouraging and supporting this research as well.
The research benefitted from the generous cooperation of the College Board in
providing us access to the SAT data files. Wayne Camara and Howard Everson
were especially helpful in expediting the process. We want to also thank the
National Center for Education Statistics for conducting analyses on the
Baccalaureate and Beyond database to provide SAT performance for college
graduates.
Finally, a number of key individuals contributed mightily to this effort. At
ETS, Norma Norris was responsible for organizing the data and running the
analyses; her counterparts at ACT were Dina Bassiri and Kevin Andrews.
Ren Lawless organized the final report and made all our data presentable.
Christian Hilland served as general assistant to the project.
We also would like to thank several individuals who reviewed drafts of this
report and helped us clarify our thinking, including Paul Barton, Rich Coley,
Jane Faggen, Art Wise, Barbara Bruschi, Janet Spiegel, and others mentioned
already.
Of course, any errors or lack of clarity are the responsibility of the authors
alone.

ABSTRACT
This study examined the academic and demographic profile of the pool of
prospective teachers and then explored how this profile is affected by teacher
testing. Specifically, the study linked SAT and ACT college admissions test data
from 1977-1995 with data from more than 300,000 prospective teachers who
took a college of education entrance exam or teacher licensure test from The
Praxis Series between 1994 and 1997. The data revealed that teacher
academic ability varies widely by type of licensure sought, with those candidates
seeking licenses in academic subject areas having the highest college admissions
test scores, and those in non-academic fields like elementary education having
the lowest scores. In contrast to many previous research claims that teachers
lack the academic ability of other college-educated professionals, the data in this
study suggest that teachers in academic subject areas have academic skills that
are equal to or higher than those of the larger college graduate population.
Across the board, teacher testing was found to positively influence the average
SAT and ACT scores of the prospective teacher pool, while at the same time
limiting the overall supply of teachers. The data further revealed that the
prospective teacher pool is highly homogenous with respect to race/ethnicity,
and that disparate passing rates on teacher tests limit the racial/ethnic diversity
of the teaching force even further. If minimum passing scores on teacher tests
are raised, as many advocates of high standards have recommended, the SAT
and ACT scores of the prospective teacher pool will rise dramatically, but the
supply and diversity of the pool will fall equally dramatically. The authors
conclude that teacher testing holds great promise, but must be used judiciously
and in combination with other reform efforts to ensure an adequate supply of
academically talented and racially/ethnically diverse teachers.

THE CONTEXT: TEACHER ACADEMIC ABILITY, SUPPLY,

AND DIVERSITY
The educational reform movements of the 1980s and 1990s have placed
Americas teachers under a microscope. Numerous blue-ribbon panels have
characterized the American educational system as woefully inadequate to meet
the demands of the 21st century (Carnegie Forum on Education and the
Economy, 1986; National Commission on Excellence in Education, 1983; The
Holmes Group, 1986); in many cases, these panels have placed the blame
largely on Americas teachers and the institutions that train them. Typically,
reformers charge that teaching does not attract high caliber students, and argue
for higher academic standards for pre-service teachers, including more selective
entrance requirements for colleges of education, more rigorous academic
coursework for education majors, and more challenging licensure requirements
for individuals seeking entry to the profession. But are teachers really less
academically capable than other professionals? Will raising undergraduate
academic standards for prospective teachers have enough of a positive impact
on the educational system to outweigh any detrimental side effects? How
successful are teacher education institutions in preparing potential teachers?
Although these questions have been asked for many years, the answers remain
the subject of debate and controversy. This report seeks to inform the debate
by profiling the academic and demographic characteristics of prospective
teachers, and by analyzing the impact of testing on the prospective teacher
pool.
Researchers have expressed serious concerns about the academic ability of
teachers since at least the 1920s (Carnegie Forum on Education and the
Economy, 1986; Ekstrom & Goertz, 1985, March; Haney, Madaus, & Kreitzer,
1987; Kerr, 1983; Koerner, 1963; Lanier, 1986; Lee, 1984; Sowell, 1993;
Weaver, 1983). And year after year, high school seniors who intend to major in
education have earned lower scores on college admissions tests of verbal and
quantitative ability than their college bound peers (ACT, 1997; The College
Board, 1997). This trend is particularly troubling in light of a growing body of
evidence establishing a link between teacher verbal ability, as measured on
standardized tests, and their studentsachievement on standardized tests
(Coleman et al., 1966; Ehrenberg & Brewer, 1995; Ferguson, 1991; Greenwald,
Hedges, & Laine, 1996; Hedges & Greenwald, 1996; Strauss & Sawyer, 1986).

Concerns over teacher academic ability have become more critical as the
cognitive challenges and expectations for teachers have risen. Todays
knowledge-based economy makes effective education for all more pressing
than in the past (Darling-Hammond & Cobb, 1996). Moreover, the American
K-12 student population is more diverse with respect to income, social class,
race/ethnicity, native language, and learning needs than at any other time in
history. The democratic ideal holds that all students deserve equal educational
opportunities and experiences; teaching such a diverse array of students
requires teachers to draw upon a vast repertoire of instructional strategies and a
strong foundation of content knowledge.
Academic ability is only one of several troubling issues being raised about our
nations teachers. Foremost among other concerns is the growing demand for
teachers, brought about by increases in the student population (National Center
for Education Statistics, 1998b), turnover in the teaching force (National
Association of State Boards of Education, 1998; National Center for Education
Statistics, 1997b; National Commission on Teaching and America's Future,
1997; U.S. Department of Education, 1998), and legislation decreasing class
sizes (Hirsch, Koppich, & Knapp, 1998). Most predictions place the demand
for new hires above 2 million over the next decade, or more than 200,000
teachers per year. This demand comes at a time when colleges of education
typically confer fewer than 110,000 undergraduate degrees per year (National
Center for Education Statistics, 1998a).
Where will these new teachers come from? States have frequently responded to
increased demand by lowering standards through measures like emergency
certification, thus illustrating the conflicts inherent in trying to raise the
standards for teachersacademic performance and increase supply
simultaneously (Boe & Gilford, 1992; National Center for Education Statistics,
1995). Emergency credentials boost supply by allowing candidates to bypass, at
least temporarily, licensure requirements designed to help ensure a high-quality
teaching force. Darling-Hammond and Cobb (1996) noted that 46 states grant
emergency licenses to untrained applicants and that more than 30 officially
sanction alternative routes to licensure, some of which require only a few
weeks of training prior to entering the classroom. In 1994, 21% of newly hired
public secondary school teachers did not even have an undergraduate minor in
their primary teaching field, and more than 20% were practicing with either a
substandard license or no license at all in their primary teaching field (The
National Commission on Teaching & America's Future, 1997).
Not only do states issue emergency licenses as loopholes into teaching, they
sometimes lower the teacher training and licensing standards themselves.
Facing the prospect of having to hire 300,000 teachers over the next decade,
California recently bucked a national trend by signing into law a measure
allowing its teacher education programs to ease the requirements for fifth year
programs. (Fox, 1998). Albert Shanker (1996), former president of the
American Federation of Teachers, characterized the passing scores established
by many states on their college of education entrance exams and teacher
licensure tests as ridiculously low (p. 222) and sufficient only to keep
illiterates (p. 221) out of teaching. Presumably, at least part of the reason
states do not impose higher passing scores is for fear of lessening the supply.

A few states have been able to establish and maintain high standards, however.
The Connecticut Beginning Educator Support and Training (BEST) portfolio
system has been touted as an example of a rigorous performance assessment
that has helped lead to a more academically able teaching force in Connecticut,
as demonstrated by higher SAT scores (U.S. Department of Education, 1998).
But Connecticuts high standards do not come cheaply. The state is able to
maintain teacher supply in large part because it has by far the highest average
teaching salary in the nation $51,495 in 1994-95 (National Center for
Education Statistics, 1996). Because most states are not willing to pay such high
salaries, the far more common response to the demand for more teachers is to
lower standards.
Lowered standards have helped to ensure that, despite the huge demand for
teachers, fewer than 1% of all teaching positions go unfilled (National Center
for Education Statistics, 1997a). But while some have argued that there is no
general teacher shortage, almost all agree that there are, and will continue to be,
shortages in specific locations and subject areas (Ballou, 1996). These shortages
are far more likely to occur in poor schools with large percentages of minority
students (Darling-Hammond, 1998). Poor urban and rural schools find it more
difficult to recruit and retain well-qualified teachers, making them more likely
than affluent suburban schools to resort to lowering their standards for hiring
new teachers. And, since poor urban schools typically have disproportionately
high populations of minority students, the teacher crisis threatens the
democratic ideal of equal opportunity regardless of racial/ethnic background.
Not only do students in poor schools face a tough climb out of poverty, they
face it with teachers who are the least well prepared academically. Thus, the
real issues surrounding teacher supply and demand are not so much about
quantity as they are about quality, distribution, and equity (Hirsch et al., 1998).
Increased demand is further complicated by a lack of demographic diversity
within the teaching force. The ratio of majority to minority students is steadily
decreasing, and forecasts call for this trend to continue in the decades ahead.
While minorities made up one third of the student population in 1993-1994,
nearly 9 out of 10 public school teachers are White, and approximately 3 out of
4 are female (American Association of Colleges for Teacher Education, in
press). Nor does this profile appear likely to change significantly any time in
the near future, as 86% of teachers in their 20s are White, and 84% are female
(Feistritzer, 1996).

The Impact of
Teacher
Testing

Typically, the push to diversify the race/ethnicity of the teaching force has
collided head-on with the higher standards movement. Standards have
frequently been enforced through teacher tests. Yet minority candidates have
traditionally scored less well than their majority peers on such standardized
tests (Darling-Hammond, Dilworth & Bullmaster, 1996; Ekstrom & Goertz,
1985, March; Goertz & Pitcher, 1985; Murnane, 1991; Murnane et al., 1991;
Schlechty & Vance, 1983; Sykes, 1983; Toch, 1991; Zimpher & Yessayan,
1987). Many critics of teacher testing fear that raising standards will mean
raising the minimum acceptable scores on such tests, thereby excluding a
disproportionate number of minorities from the profession. Of course, it is the
right of each state to individually set its own testing requirements, including the
kinds of tests required and the associated passing scores. But the picture has
often been painted as a dichotomous one, in which states must decide between
raising standards or increasing diversity, with the decision in favor of one
necessarily coming at the expense of the other (Smith, 1987).
Finding the optimal balance among teacher academic ability, supply, and
diversity has thus proven treacherous, and of all the reforms proposed, perhaps
none has generated more controversy than teacher testing. Such testing has
grown explosively since the 1980s, largely because policymakers have embraced
it as the most effective means for ensuring an academically able teaching force
(National Association of State Boards of Education, 1998). Today, virtually all
states have some sort of teacher testing requirement, whether for entry into a
college of education, to earn an initial license to practice, or both (National
Association of State Directors of Teacher Education and Certification, 1998).
Yet, at the same time, such testing has been blamed for restricting both the
supply and racial/ethnic diversity of the teaching force. The true effects of
teacher testing have thus become the focus of an ongoing debate that to date
has spurred a great deal of rhetoric, but as yet little consensus (Boe & Gilford,
1992; Darling-Hammond, 1998; Darling-Hammond & Cobb, 1996; DarlingHammond & Wise, 1983; Goertz et al., 1984; Ingersoll, 1996; National
Commission on Teaching & Americas Future, 1996; Sykes, 1983).

Reforms
Within
Colleges of
Education

Colleges of education traditionally have endured scathing criticism by critics

who have charged that because education programs admit the least able college
students, out of necessity education curricula lack academic rigor (Koerner,
1963; Lanier, 1986). Approximately 30 states have attempted to address this
concern by legislating that students must pass admissions tests prior to entering
teacher education programs. Virtually all these tests focus on basic skills, as
measured by assessments of reading, writing, and mathematics ability (National
Association of State Directors of Teacher Education and Certification, 1998).
But while there is widespread consensus that education programs do not have
sufficiently rigorous entrance requirements, the impact of raising admissions
standards has not yet been analyzed systematically.

The teacher education curriculum, too, has come under heavy fire. It has
alternately been characterized as laden with too much theory and not enough
practice (Riley, 1998), as theory disconnected from practice (Shanker, 1996),
and as subject matter disconnected from teaching methods and learning theory
(National Association of State Boards of Education, 1998). Harsh criticism has
been directed in particular at the perceived mediocrity of subject-area content
training within an education major (Galambos, 1985). Few would dispute that
teachers should have a strong enough foundation of knowledge in their subject
areas to enable them to be flexible and responsive in their instruction
(Grossman, Wilson, & Shulman, 1989; Howey, 1996; Reynolds, 1992). While
subject area knowledge is considered essential for effective pedagogy, however,
by no means is it considered sufficient. Indeed, Shulman (1987), while
acknowledging that teachers must have a firm knowledge of subject matter,
identified pedagogical content knowledge as that special amalgam of content
and pedagogy that is uniquely the province of teachers (p. 8).
Today, calls to abolish the undergraduate education major in favor of content
area majors have the explicit support of the U.S. Secretary of Education (Riley,
1998). Already, 300 of the more than 1,200 colleges of education have
instituted extended teacher training programs, many of which allow graduates
to receive an undergraduate degree in their academic field and then earn a
masters in teaching during a fifth year of schooling that focuses largely on
clinical experiences within a structured, supportive environment (DarlingHammond, 1998).
Accreditation of teacher education programs has been alternately lauded and
derided as another means for improving teaching. The National Council for
Accreditation of Teacher Education (NCATE), in existence since 1954, serves
as the preeminent national accreditation body in teacher education. NCATE
has claimed that it benefits education by: (1) ensuring that colleges of education
meet external quality standards; (2) encouraging institutions to modify their
programs to reflect changes in knowledge and practice; (3) providing a
common set of national standards; (4) strengthening institutional self-evaluation
and catalyzing program improvement; and (5) deterring decreases in resource
allocations (Roth, 1996). Proponents have asserted that NCATE has led the
way in changing teacher preparation to match more rigorous licensing and
master teacher certification requirements, and in encouraging links between
student and teacher standards (Wise & Leibbrand, 1996). Today, NCATE
accredits approximately 500 of the 1,200 colleges of education. These colleges
account for approximately two thirds of all education graduates annually; many
of the 700 unaccredited institutions graduate very small numbers of teacher
candidates (Wise, 1997). Accreditation, along with more comprehensive
teacher licensure (e.g., Interstate New Teacher Assessment and Support
Consortium) and advanced certification (e.g., National Board for Professional
Teaching Standards), form the three-legged stool of teacher qualityadvocated
by the National Commission on Teaching and Americas Future (1996, 1997).
Since its inception, however, NCATE has proven controversial (Gardner,
Scannell, & Wisniewski, 1996). While some institutions have credited NCATE
with turning their program around, others have argued that the requirements
are too prescriptive and unrepresentative (Gardner et al., 1996; Nicklin, 1992),
and that NCATE institutions as a whole do not have sufficiently rigorous
admissions requirements (Ballou & Podgursky, 1998).

THE STUDY
Key Issues

Our objective is to inform the policy debates that continue to swirl around the
critical issues of teacher academic preparation, supply and demand, diversity,
education programs, and program accreditation. We do this by drawing on
available data from those individuals who have taken both a teacher test from
The Praxis Series and either of the two major college admissions tests, the
SAT or the ACT. We address the following key issues:
What are the academic and demographic characteristics of the prospective teacher pool,
and how does teacher testing impact this pool?
Though many claims about the academic quality of teachers are based on SAT
and ACT data collected from high school students intending to major in
education, these scores are a flawed proxy for the academic ability of the
teaching force. Hanushek and Pace (1994) estimated that more than half of all
potential education majors switch career plans between their senior year in high
school and the end of their sophomore year in college. Moreover, about one
quarter of the teaching force did not major in education (Feistritzer, 1996;
NCES, 1996), and this number is likely to grow as current initiatives supporting
subject area majors and alternate routes to teaching take hold.
One way around these problems is to look at the SAT and ACT scores, as well
as undergraduate grades, of those who actually seek to become licensed
teachers. We identified all individuals who took a test from Educational
Testing Services (ETSs) Praxis Series between Fall 1994 and Spring 1997. The
Praxis Series is administered in 34 states and the District of Columbia either
for entrance to college of education programs, for teacher licensure, or for
both.1 We then searched SAT and ACT records back through 1977 and
created a matched data set that enabled us to provide a comprehensive look at
those who are well into the teacher pipeline, particularly in contrast to potential
education majors. The critical data are not the academic and demographic
characteristics of a pool of 18-year-olds who think they might major in
education, but rather the characteristics of those who are actually pursuing
teaching upon graduating from college.
We must be able to differentiate the qualities of individuals who seek licensure
from those who actually qualify for licensure. If teacher education programs
are effective, then it is quite possible that through entrance requirements,
program requirements, and licensure testing, some significant number of
potential teachers will not become licensed to teach. In this study, we
systematically ask two questions:

What are the academic and demographic characteristics of all

those who seek licensure?

What are the academic and demographic characteristics of those

who actually pass the licensure test requirements? In other words,
how does licensure testing affect the pool of potential teachers?

See Appendix A for a complete listing of the Praxis user states in 1997.

We also examine analogous issues for admission into teacher education

programs, namely:
What are the academic and demographic characteristics of all
those who take teacher education entrance examinations?
What are the academic and demographic characteristics of those
who actually pass the entrance test requirements? In other words,
how does program entry testing affect the pool of potential
teaching students?
We also compare the academic characteristics of prospective teachers, as
measured by SAT and ACT scores, with appropriate contrast groups. We first
ask, how do those who pass college of education entrance tests compare with
the pool of college bound high school seniors? Do the results look different
from simply examining the data for intended college education majors?
We then ask, how do those who pass teacher licensure tests compare with all
those who graduate from college? Existing data, though limited, suggest that
college admissions test scores of college graduates may be significantly higher
than for all those who enter college (Hanushek & Pace, 1995). The most
appropriate comparison group for teachers virtually all of whom are college
graduates (National Center for Education Statistics, 1999) is obviously others
who have also completed college.
What is the relationship between academic quality and licensure area?
Claims about teaching are often stated as if the teaching force were a
homogeneous group. Yet academic requirements for most elementary teachers
are typically far different from those for teachers of mathematics, for example.
In this study, we compare the academic profiles of those who seek different
kinds of teaching licenses and then compare these results with those of all
college bound seniors, intended education majors, and college graduates.
How would raising licensure testing standards affect the academic and demographic
profiles of the prospective teaching population?
We explore several hypothetical questions in this study. While we analyze most
of our data in terms of established passing scores for each state, we also
investigate what would happen to the prospective teaching population if we set
a single passing score for all candidates across all states. What would happen to
the academic and demographic profiles of the prospective teaching population
if we set a uniformly high passing standard, and correspondingly, what would
happen if we set a uniformly low passing standard?

10

 What is the relationship among teacher education training, teacher program accreditation,
and the academic and demographic profiles of the prospective teaching population?
We examine academic characteristics and performance on licensing tests as a
consequence of attending teacher education institutions in contrast to preparing
for licensure through alternate pathways. We also examine the characteristics
and licensing performance of those who attend NCATE institutions as
compared with those who attend institutions not accredited by NCATE. We
further contrast the characteristics of those who major in general education
programs with those who major in a content area.

Study
Overview

The Praxis Series is the only national teacher-testing program currently in

operation. Praxis I tests assess reading, writing, and math ability, and are
generally required for entrance into a college of education. Praxis II tests focus
on content and pedagogical knowledge in specific subject areas, and are used by
states to grant initial teaching licenses. Praxis II also offers several series of
licensing tests that focus on more generic teaching knowledge and pedagogical
skills.
Almost 600,000 Praxis tests were administered during the three-year window
examined in this study. Each teacher candidate who tested completed a
background information questionnaire that asked for information on
race/ethnicity, age, gender, parentseducation level, language spoken at home,
undergraduate and graduate majors, highest degree attained, college attended,
and other variables. For all candidates, we considered the data listed on the
candidates most recent Praxis test registration form.
We created two parallel data sets: one for candidates who took the SAT and
one for candidates who took the ACT. We searched both the SAT and ACT
data sets from 1977 to 1995. We then matched the most recent SAT/ACT
scores with the Praxis scores and background information that came from both
the Praxis questionnaire and the SAT and ACT background questionnaires.
For the SAT, math and verbal scores were included.2 For the ACT, English
and math scores were included. After completing the matches, there were
33,866 Praxis I candidates and 159,857 Praxis II candidates who had taken the
SAT, and 55,064 Praxis I candidates and 112,207 Praxis II candidates who had
taken the ACT. Data were analyzed separately for Praxis I and II. Individuals
who took Praxis I and Praxis II tests between 1994 and 1997 appear in both
Praxis data sets.
A third match was done to determine whether candidates had attended an
NCATE-accredited college as undergraduates.3 These data were then matched
to the Praxis and SAT/ACT database through the attending institution
information provided by candidates when they registered to take Praxis tests.

All scores are reported on the SAT scales that were recentered in 1995.
The American Association of Colleges for Teacher Education (AACTE) generously supplied an electronic file
containing a master list of undergraduate institutions in the United States that were affiliated with NCATE as of 1997.
Note that a small proportion of these institutions (approximately 8%) were in the process of accreditation, but were not
fully accredited by NCATE in 1997. Because the vast majority of these schools would eventually receive accreditation,
our analyses included these institutions as well.
2
3

11

We then assigned a state passing status to each candidate. For each Praxis test
taker, we applied the passing scores in effect within the state in which the
candidate tested. In each case, we applied the respective states passing
standard in 1997, even though a state may have changed its passing standard
between 1994 and 1997.
Further, we had to define what it meant for a candidate to be considered
passing. In many cases, candidates must take multiple Praxis II tests to
become licensed; for Praxis I candidates, multiple tests are almost always
required. Candidates were considered to have passed if they passed all tests
they had taken in a licensure area or in the Praxis I battery. So, for example, a
candidate who took two Praxis II mathematics tests had to pass both in order
to be considered licensed in mathematics. A candidate who did not meet the
passing standard on one or both tests was considered not to have met the
licensing requirement. A candidate who passed all tests taken in a certification
area was considered to be a pass even if the state required more tests than the
candidate had taken during the three-year window for which we had data.
Since we could not determine how well candidates did on tests either prior to
or subsequent to this three-year window, we based our estimation of passing
status only on tests taken.
For each candidate, we also assigned passing status based on two hypothetical
passing standards, low and high. Passing standards vary substantially among
states for most of the Praxis tests. We set the high standard for every test at
the highest passing score used by any state in 1997, and the low standard at the
lowest passing score used by any state in 1997. We then considered each
candidates passing status in light of the test requirements of the state and these
hypothetical passing standards. For all analyses, the most recent Praxis test
scores were used in those instances when a candidate took the same test
multiple times.
For the majority of the analyses, we used the state passing status as the basis for
comparisons. For each variable of interest, we analyzed the makeup of the
candidate pool, including SAT/ACT scores, and then compared those who met
Praxis passing requirements with those who did not. We also conducted a
series of analyses using the hypothetical low and high passing scores as the
basis for understanding the potential impact of raising or lowering teacher
testing standards.

Limitations
and Caveats

A study of this type is bound to have certain limitations that place restrictions
on the interpretation of its findings, and it is critical not to overgeneralize the
results from this or other related studies.
Perhaps the most significant limitation has to do with our defining academic
ability through SAT/ACT scores. Obviously, these standardized test scores
present a narrow picture of an individuals academic skills, and there are
unquestionably many more facets to academic abilitythan SAT/ACT scores.
Nevertheless, we know of no other widely available data that enable
trustworthy comparisons of individualsacademic qualities.

12

SAT/ACT scores do provide objective measures of skills valued by many

colleges. But why consider SAT/ACT scores at all if they do not purport to
predict teacher effectiveness? We do not mean to imply in any way that
candidates who perform well on the SAT or ACT will automatically make good
teachers, nor that someone who performs poorly on the SAT or ACT cannot
excel as a teacher. Nonetheless, at least two lines of reasoning support the
appropriateness of studying the SAT/ACT scores of teachers. The first is that
if schooling is to be considered an academic enterprise, then it seems only
logical that teachers be drawn from among the more academically able; all
things being equal, academic ability is clearly a desirable trait in teachers. A
second comes from growing evidence that verbal ability, as measured in the
standardized test scores of teachers, is positively related to studentstest scores
(Ehrenberg & Brewer, 1995; Ferguson, 1998). So while SAT/ACT scores
provide an incomplete proxy for academic ability, and many qualities unrelated
to academic ability go into making an accomplished teacher, It would be
absurd to argue that academic ability is not or should not be at least one
measure of teacher quality (Weaver, 1983, p. 1).
A second limitation of the study is that the Praxis database is not completely
representative of the entire teacher candidate population. Not all states use The
Praxis Series , and in a small number of states that do, not all teacher
candidates are required to take each of the tests offered. For example,
completion of specific teacher education programs in California may exempt an
individual from taking licensing tests.
Moreover, the database was created by taking all Praxis I and II test takers from
1994 to 1997, and then matching them by social security number to all
candidates who took the SAT/ACT during the previous 20 years. Teacher
candidates who did not take both a Praxis test and the SAT or ACT in the time
periods specified were dropped from the database. Those who took the SAT
or ACT prior to 1977, for example, would not appear in the database. To be
accurate, these data represent only those teacher candidates who took both the
SAT or ACT and a Praxis test within a prescribed time period. Still, the study
includes more than 300,000 candidates who were in the teaching pipeline
between 1994 and 1997, and there is no compelling reason to believe that the
samples overall profile is skewed substantially with respect to that of the
overall prospective teaching population.
Another limitation of the study is that all background information was selfreported, and self-reporting may introduce bias. For example, the analyses by
race/ethnicity would be skewed if candidates from one racial/ethnic group
were less likely than others to identify their racial/ethnic background on the
questionnaire. But, since there was no clear way to identify erroneous or biased
background data, no adjustments were made to the self-reported information in
the database.
Finally, in considering the study data, it is important to bear in mind that Praxis
tests are not designed to predict teacher effectiveness. As program entrance
and licensure tests, they measure knowledge considered essential to effective
pedagogy, but do not attempt to measure the full breadth of skills that go into
being an accomplished teacher. Therefore, passing a Praxis test does not
guarantee that an individual will become a satisfactory teacher. It does,
however, warrant that the individual has acquired a level of knowledge that is
adequate for a beginning teacher.

13

STUDY FINDINGS
Given the complex interrelationships among teacher academic ability, supply,
and diversity, combined with the debate and controversy surrounding the role
of teacher testing in the reform movement, it is critical to provide hard data
about the impact of testing on the pool of potential teachers. This section
presents the results of our analyses. In most cases, the SAT and ACT data are
consistent with one another, and therefore, the accompanying discussion
usually considers them together. In the relatively rare instances when the SAT
and ACT data present different pictures, the discussion explicitly addresses
these differences.
We typically discuss the data on two levels: for the overall population taking the
Praxis tests, and for those who pass in comparison with those who do not. We
chose this strategy because it is important to understand who is seeking entry to
the teaching pipeline, both at the undergraduate and licensure stages, and how
testing affects the pool of potential teachers. Those who pass Praxis tests are
permitted to continue further in the pipeline, while those who fail cannot
proceed until they pass.

SAT/ACT
Baseline Data

Table 1:

SAT scores are reported on math and verbal scales that range from 200 to 800,
and the corresponding ACT math and English scales range from 1 to 36. The
data in Table 1 present average SAT and ACT scores by gender and
race/ethnicity for high school students in 1997. These data are intended to
provide a context for understanding the subsequent discussion of SAT and
ACT scores for various groups of Praxis test takers.

Mean SAT and ACT Performance by Gender and Race/Ethnicity

SAT

ACT

Math
511

Verbal
505

Math
20.6

English
20.3

Male
Female

530
494

507
503

21.3
20.1

19.9
20.7

African American
Asian American/Asian
Hispanic
Native American
White
Other

423
560
458
475
526
514

434
496
457
475
526
512

17.8
24.1
20.2
20.0
22.3
20.5

17.4
21.2
19.1
19.5
22.2
19.9

All Test Takers

Sources: The College Board. (1997). College Bound Seniors: A Profile of SAT Program Test Takers. New York: Author.
ACT. (1997). The High School Profile Report: Normative Data. Iowa City, Iowa: Author.

14

Average SAT and ACT scores tend to remain relatively stable from year to year.
For example, for the years covered in this study, 1977-1995, the average SAT
math scores ranged from 492 to 506, and the average SAT verbal scores ranged
from 499 to 507 (The College Board, 1997); with the introduction of the
Enhanced ACT in 1989-1990, mean mathematics and English test scores have
ranged from 19.9 to 20.8, and 20.2 to 20.5, respectively (ACT, 1997).
Math SAT scores do vary considerably by gender, however. This fact is
especially important to keep in mind when one considers that the overall
teaching population is nearly 75% female. Because, on average, females do not
do as well as males on the composite SAT, we would expect a predominantly
female sample (e.g., teachers) to have a lower SAT profile than a random
sample of individuals that included equal numbers of males and females.

Results For
Those
Entering
Teacher
Education
Programs

Table 2:

Table 2 provides further context for interpreting the Praxis I data. Praxis I tests
typically are used for determining whether or not a candidate can enroll in a
college of education.

Mean SAT/ACT Scores for Praxis I Comparison Groups

SAT

ACT

Intended Education Majors

Math
479

Verbal
485

Math
19.6

English
19.8

All College Bound Seniors

Male
Female

511
530
494

505
507
503

20.6
21.3
20.1

20.3
19.9
20.7

Praxis I Test Takers

491

503

19.9

20.9

Praxis I Candidates Who Pass

Male
Female

514
535
507

525
527
524

20.4
21.2
20.1

21.6
20.8
21.8

Sources: The College Board. (1997). College Bound Seniors: A Profile of SAT Program Test Takers. New York: Author.
ACT. (1997). The High School Profile Report: Normative Data. Iowa City, Iowa: Author.

15

Several interesting trends emerge from the data in Table 2. First, the data
clearly demonstrate the oft-cited disparity between intended education majors
and their college bound peers with respect to admissions test scores. But it
would be remiss to use these data alone as evidence that education majors trail
their peers in academic ability, because the scores for Praxis I candidates are
higher than those of high school students expressing an interest in education.
Those who pass Praxis I tests have even higher scores, scores that are
comparable to those of all college bound seniors in math, and substantially
higher than all college bound seniorsverbal/English scores. The relative
strength of the Praxis I passing populations scores becomes even more
apparent when the results are broken out by gender, thereby controlling for the
relatively large percentage of females in the Praxis I population. For both
males and females, successful Praxis I candidates have composite college
admissions scores that are higher than those of the pool of college bound
seniors.
Because the Praxis I population consists primarily of college students actively
seeking entrance to a college of education, the Praxis I passing population
clearly provides a better proxy for eventual education majors than does the
population of high school seniors intending to major in education. The data in
Table 2 suggest that, contrary to assertions in much of the research literature,
students entering colleges of education have test scores that are comparable to
or even higher than those of the larger college bound population.
Table 3 presents passing data for Praxis I test takers overall and by gender.
Results are given separately for Praxis/SAT test takers and Praxis/ACT test
takers. Almost all subsequent tables in this report will contain the same
columns shown in Table 3: % Passrepresents the percentage of candidates
within each subgroup who passed the test; Nstands for the total number of
test takers in each group; and % Pooldefines the composition of the
candidate pool. Table 3 shows, for example, that 79% of the males and 77% of
the females who took both Praxis I and the SAT met the passing standard for
Praxis I. Of the 33,866 people who took both Praxis I and the SAT, 26,182
passed Praxis I. The pool of passers was 75% female and 25% male. Those
who passed had average math and verbal SAT scores of 514 and 525,
respectively, compared to average scores of just 414 and 427, respectively, for
those who did not pass. ACT data are presented using the same reporting
structure.

16

Table 3:

Mean SAT/ACT Scores by Passing Status and Gender (Praxis I)

All Praxis I Test Takers

All Candidates
Male
Female

%
Pass

N4

SAT
% Pool Math Verbal

33,866 -8,242 24%

25,624 76%

491
513
484

503
505
502

77%
79%
77%

26,182 -6,537 25%

19,645 75%

514
535
507

525
527
524

Candidates Who Do Not Pass

All Candidates
Male
Female

7,684 -1,705 22%

5,979 78%

414
430
409

427
420
429

Candidates Who Pass

All Candidates
Male
Female

% Pass

88%
89%
87%

ACT
% Pool Math English

55,064 -14,833 27%

40,231 73%

19.9
20.7
19.6

20.9
20.1
21.2

48,248 -13,164 27%

35,084 73%

20.4
21.2
20.1

21.6
20.8
21.8

6,816 -1,669 24%

5,147 76%

16.5
17.0
16.4

16.1
15.2
16.4

The number of ACT takers who took Praxis I is considerably higher than the
number of SAT takers who took Praxis I, even though the SAT has higher
annual administration volumes. The reason for this is likely because Praxis I is
required for all students by institutions and/or licensing agencies in a number
of Midwestern states, and the ACT is taken more frequently than the SAT by
students in the Midwest.
One notable discrepancy between the SAT and ACT data is that nearly 9 in 10
ACT takers passed Praxis I, as opposed to under 8 in 10 SAT takers. The
reason for this disparity is not immediately clear, but is likely attributable to a
combination of factors, including candidate demographics.
Approximately three fourths of the Praxis I candidates are female. Males and
females pass at about the same rate, so Praxis I tests do not influence the
composition of the prospective teaching pool with respect to gender.
SAT/ACT scores are much higher for those who pass Praxis I than for those
who do not.
Table 4 presents the Praxis I data by race/ethnicity. The composition of the
pool is consistent with portrayals of the teaching force as racially/ethnically
homogeneous, particularly with respect to the ACT population, in which 90%
of the candidates are White, just 5% are African American, and other minority
candidates combined make up less than 5% of the pool. The SAT population
is more diverse, with substantially higher percentages of African American and
Asian American/Asian candidates, but even so, more than 8 in 10 of the SAT
candidates are White.
The numbers of candidates reported in this and subsequent tables represents only those candidates who self-reported
the relevant background information on the Praxis biographical questionnaire. Thus, the total number of candidates
varies from table to table.
4

17

Table 4:

Mean SAT/ACT Scores by Passing Status and Race/Ethnicity (Praxis I)

All Praxis I Test Takers

All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other
Candidates Who Pass
All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

%
Pass

SAT
%
Pool

Math Verbal

33,770
3,603
1,277
602
187
27,506
595

-11%
4%
2%
1%
81%
2%

491
413
517
459
457
501
488

503
428
490
476
459
514
507

77% 26,115
46% 1,650
76%
966
69%
418
64%
120
82% 22,537
71%
424

-6%
4%
2%
0%
86%
2%

514
456
542
490
490
517
526

525
474
517
506
497
529
547

-26%
4%
2%
1%
65%
2%

414
378
440
388
397
428
395

427
389
406
408
391
445
409

Candidates Who Do Not Pass

All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

7,655
1,953
311
184
67
4,969
171

% Pass

88%
63%
77%
83%
81%
89%
84%

ACT
% Pool Math English

54,797
2,829
728
694
555
49,548
443

-5%
1%
1%
1%
90%
1%

19.9
16.4
20.8
18.6
17.4
20.1
19.9

20.9
17.0
19.6
19.3
18.1
21.2
20.7

48,036
1,790
558
575
450
44,293
370

-4%
1%
1%
1%
92%
1%

20.4
17.2
21.7
19.2
17.8
20.5
20.6

21.5
18.4
21.1
20.1
18.9
21.7
21.6

6,761
1,039
170
119
105
5,255
73

-15%
3%
2%
2%
78%
1%

16.5
15.0
17.4
15.8
15.5
16.9
16.0

16.1
14.5
14.8
15.7
14.9
16.5
15.9

Praxis passing rates differ substantially by race/ethnicity, with White candidates

passing at the highest rate, and African Americans at the lowest. These
disparate passing rates serve to limit the racial/ethnic diversity of the teaching
force; however, it is worth noting that the overall Praxis I population lacks
diversity even when passing status is taken out of the equation.
The relationship between SAT/ACT scores and passing rates is fairly
consistent, as groups with higher standardized test scores also pass the licensing
tests at a higher rate. The exceptions to this relationship are Asian
American/Asian and Other candidates, who tend to have college admissions
test scores that are higher than White candidates,but whose passing rates are
significantly lower. Across all groups, those who pass Praxis I have
substantially higher SAT/ACT scores than those who do not.

18

In sum, Praxis I appears to have precisely the impact that both proponents and
detractors of teacher testing claim it has: a significant positive impact on the
academic profile of the prospective pool of education majors, at the price of
limiting supply and racial/ethnic diversity.

Results For
Those Who
Seek
Licensure

We also conducted analyses of candidates who took the Praxis II assessments,

which are used as part of the licensing requirements in most states. We found
that, overall, SAT/ACT scores for those who passed the licensing tests were
comparable to those who succeeded on the Praxis I assessments. As with
Praxis I, we compared the SAT scores of Praxis II test takers with those of
college bound seniors. But because Praxis II candidates have completed all or
most of their college career, it is more accurate to compare their scores with
those of all college graduates. Individuals with relatively low SAT/ACT scores
are less likely to complete college than their higher scoring peers, so SAT/ACT
scores for college graduates are likely, because of attrition, to be higher than the
scores of college aspirants. Note however, that the relative number of test
takers demonstrates that most Praxis II test takers were not required to take the
Praxis I tests. The majority of Praxis II candidates entered teacher education
programs by satisfying requirements other than the Praxis I assessments.
Table 5 shows that composite SAT and ACT scores for all Praxis II test takers
are similar to the respective scores for all college bound seniors. However,
composite scores for those who pass the Praxis II exams are slightly higher
than for college bound seniors. As with the Praxis I data, disaggregating by
gender clarifies that, indeed, both male and female Praxis II test takers have
SAT/ACT scores that are comparable to or higher than those of the pool of
college bound seniors.

Table 5:

Mean SAT/ACT Scores for Praxis II Comparison Groups

SAT

ACT

Math
511
530
494

Verbal
505
507
503

Math
20.6
21.3
20.1

English
20.3
19.9
20.7

All College Graduates

542

543

--

--

Praxis II Test Takers

498

511

19.7

21.5

Praxis II Candidates Who Pass

Male
Female

507
529
500

522
525
521

20.1
20.8
19.9

22.0
21.1
22.3

All College Bound Seniors

Male
Female

Sources: The College Board. (1997). College bound seniors: A Profile of SAT program test takers. New York: Author.
ACT. (1997). The High School Profile Report: Normative Data. Iowa City, Iowa: Author.

National Center for Education Statistics. (1998). Baccalaureate and Beyond. Washington, DC: Author.

19

A somewhat different picture emerges when only SAT scores from college
graduates are considered.5 College graduates have much higher SAT scores
than college aspirants 31 and 38 points, respectively, on the math and verbal
scales. The math scores for Praxis II candidates are much closer to those of
college aspirants than they are to college graduates, while the verbal scores are
somewhere in between. And even though those who pass Praxis II have higher
SAT/ACT scores than the overall Praxis II population, their scores still lag well
behind those of all college graduates.
The data in Table 5 suggest that the relative standing of those in the teacher
pipeline compared with their college peers changes over time. In the career
selection process that takes place during college, the group of students who
choose teaching as a career, taken as a whole, are not as high achieving as their
college peers with respect to SAT scores.
Table 6 displays the Praxis II results by gender. The data are consistent for
both the SAT and ACT populations. Approximately three fourths of the
candidates are female. Both males and females pass Praxis II at about the same
rate, so licensing tests have virtually no effect on the gender makeup of the
pool of those who meet licensing requirements. Close to 90% of all Praxis II
test takers pass; their SAT and ACT scores are much higher than the scores for
those who do not pass

Table 6:

Mean SAT/ACT Scores by Passing Status and Gender (Praxis II)

%
Pass

All Praxis II Test Takers

All Candidates
Male
Female
Candidates Who Pass
All Candidates
Male
Female

SAT
%
Pool

Math Verbal

159,857 -40,090 25%

119,767 75%

498
520
491

511
514
509

87% 139,644 -87% 34,697 25%

88% 104,947 75%

507
529
500

522
525
521

435
464
425

434
446
430

Candidates Who Do Not Pass

All Candidates
Male
Female

20,213 -5,393 27%

14,820 73%

%
Pass

89%
87%
90%

112,207 -26,617 24%

85,590 76%

19.7
20.3
19.5

21.5
20.5
21.8

100,214 -23,274 23%

76,940 77%

20.1
20.8
19.9

22.0
21.1
22.3

11,993 -3,343 28%

8,650 72%

16.3
17.0
16.1

16.6
16.2
16.8

ACT scores for college graduates are not yet available so are not included in this analysis.

20

ACT
%
Math English
Pool

Table 7 displays the overall Praxis II results by race/ethnicity. As with Praxis I,

these data present a picture of a racially/ethnically homogeneous teaching
force. Fewer than 10% of the candidates are African American.
Representation by those with Spanish-speaking backgrounds is even lower,
approximately 2-3 % of the Praxis II pool. Similarly, few Asian Americans and
even fewer Native Americans are taking Praxis II tests. Overall, the pool of
candidates for teacher licensure is predominantly White and female.

Table 7:

Mean SAT/ACT Scores by Passing Status and Race/Ethnicity (Praxis II)

%
Pass
All Praxis II Test Takers
All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

SAT
%
Pool

Math Verbal

159,270
11,510
3,810
5,352
488
135,035
3,075

-7%
2%
3%
0%
85%
2%

498
424
534
465
478
505
497

511
440
508
473
484
518
517

Candidates Who Pass

All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

87% 139,245
69%
7,984
75%
2,874
59%
3,134
80%
388
91% 122,534
76%
2,331

-6%
2%
2%
0%
88%
2%

507
441
547
486
491
511
515

522
463
529
503
499
525
541

Candidates Who Fail

All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

20,025
3,526
936
2,218
100
12,501
744

-18%
5%
11%
1%
62%
4%

436
384
495
434
427
446
438

434
389
445
432
424
446
441

%
Pass

89%
61%
82%
81%
81%
93%
74%

ACT
%
Math English
Pool

111,591
11,111
840
2,481
631
94,846
1,682

-10%
1%
2%
1%
85%
2%

19.7
16.5
22.5
18.5
17.9
20.1
18.8

21.5
17.5
21.7
19.4
19.2
22.0
20.1

99,804
6,757
691
2,007
514
88,583
1,252

-7%
1%
2%
1%
89%
1%

20.1
17.4
23.0
18.8
18.6
20.3
19.7

22.0
19.0
22.7
20.1
20.4
22.3
21.5

11,787
4,354
149
474
117
6,263
430

-37%
1%
4%
1%
53%
4%

16.4
15.1
19.9
16.6
14.9
17.2
15.9

16.7
15.2
17.1
16.2
14.3
17.8
16.2

21

Pass rates do vary significantly by race/ethnicity. The passing rate for White
candidates exceeds 90% in both the SAT and ACT pools. The passing rate for
all other groups is significantly lower, with Hispanics and African Americans
passing the Praxis II tests at the lowest rates on the SAT and ACT, respectively.
The relationship between SAT/ACT scores and passing rates is fairly
consistent, as groups with higher standardized test scores also pass the licensing
tests at a higher rate. SAT/ACT scores invariably are much lower for
candidates, across all racial/ethnic groups, who do not pass the Praxis II
assessments. The one exception is Asian American candidates, who tend to
have composite college admissions test scores that are higher than White
candidates, but whose Praxis II passing rates are significantly lower.
The net result is that a pool of teacher candidates that is overwhelmingly White
prior to licensure testing becomes even more homogeneous after testing.
These data suggest that the most recent entrants into the teaching pool are no
more diverse, with respect to race/ethnicity, than their predecessors. The
academic profile for those who pass Praxis II, however, is stronger than the
entire candidate pool, indicating that licensure testing removes those with the
lowest SAT/ACT scores from the pool of potential teachers.
Do undergraduate grades have any meaning or, as has been suggested, has
grade inflation, particularly in the field of education, rendered them
meaningless? Table 8 shows that approximately two thirds of all teacher
candidates have self-reported GPAs of 3.0 or greater, while just 4% have GPAs
of less than 2.5. Yet, even within this restricted range, there is a consistent
relationship between GPA and licensure test passing rate. The top quartile of
students, those with GPAs of 3.5 or above, almost always pass the licensing
tests. As mean GPA decreases, so do passing rates. Still, two thirds of the C
students meet state licensing requirements.

22

Table 8:

Mean SAT/ACT Scores by Passing Status and Undergraduate GPA (Praxis II)

%
Pass
All Praxis II Test Takers
All Candidates
3.5 - 4.0
3.0 - 3.49
2.5 - 2.99
2.0 - 2.49
< 2.0

159,289
40,403
66,711
46,030
6,056
83

-25%
42%
29%
4%
0%

498
530
499
473
463
442

511
552
512
479
464
467

Candidates Who Pass

All Candidates
3.5 - 4.0
3.0 - 3.49
2.5 - 2.99
2.0 - 2.49
< 2.0

139,217
38,439
59,350
37,338
4,029
57

-28%
43%
27%
3%
0%

507
534
506
484
483
471

522
556
520
492
489
493

20,072
1,964
7,361
8,692
2,027
26

-10%
37%
43%
10%
0%

435
453
443
429
423
379

434
461
445
423
416
410

Candidates Who Fail

All Candidates
3.5 - 4.0
3.0 - 3.49
2.5 - 2.99
2.0 - 2.49
< 2.0

87%
95%
89%
81%
67%
69%

SAT
%
Pool

Math Verbal

%
Pass

89%
96%
92%
84%
70%
50%

ACT
%
Math English
Pool

111,691
28,455
44,538
33,527
4,992
168

-25%
40%
30%
4%
0%

19.7
21.6
19.8
18.3
17.4
16.5

21.5
24.0
21.5
19.6
18.5
17.8

99,801
27,438
40,765
28,020
3,487
84

-27%
41%
28%
3%
0%

20.1
21.8
20.0
18.8
18.1
17.8

22.0
24.2
22.0
20.3
19.7
20.7

11,890
1,017
3,773
5,507
1,505
84

-9%
32%
46%
13%
1%

16.4
17.7
16.6
16.1
15.7
15.2

16.7
18.9
17.2
16.2
15.6
15.0

Higher GPA is consistently associated with higher SAT/ACT scores. This

relationship holds for both those who pass and for those who do not pass the
licensing tests. Candidates with GPAs of 3.5 and above have SAT/ACT scores
that far exceed scores of all college bound seniors. In fact, the composite SAT
scores of those who pass Praxis II and have the highest grades are slightly
higher than those of all college graduates (Table 5).
Despite grade inflation, college grades do appear to retain some meaning.
Students who do better in the classroom tend to be more successful on
licensing tests. To a small extent, the teacher pool is made up of
proportionately more successful students as a result of licensure testing.
Nevertheless, the effects are small due, no doubt, to the relatively restricted
range of grades that are used.

23

Teacher
Education
Institutions

The role of teacher education institutions has come under increasing scrutiny,
with considerable debate over whether prospective teachers should major in
education or an academic field. For both SAT and ACT candidates, 86% of
those who seek licensure still come through teacher education programs. Two
thirds of candidates report that they are currently in such programs when they
take the Praxis II tests. These data are presented in Table 9.
Current teacher education students have the highest passing rate on the
licensure tests. Interestingly, those who report they have never been enrolled in
a teacher education program have the lowest passing rates, while their mean
SAT scores are as strong as the scores for those who were or had been
previously enrolled in teacher education programs when they took the licensing
tests. The ACT population of those who were never enrolled in a teacher
education program also pass at a lower rate, though their ACT scores are lower
than for students who were in teacher education programs at the time of
testing, but almost identical to those enrolled in such programs in prior years.
These results make it clear that teacher education programs have an important
impact in preparing their students to meet the requirements of licensure.
Academic ability, as measured by SAT/ACT scores does not, in itself, ensure
that someone will have the knowledge and skills that are assessed by teacher
licensure tests. Teacher education programs appear to be providing some
critical knowledge that enables education students to pass licensure tests at
higher rates than students who never enrolled in a teacher education program.

Table 9:

Mean SAT/ACT Scores by Passing Status and Enrollment in Teacher Education

Programs (Praxis II)

159,071 -104,594 66%

32,171 20%
22,306 14%

498
502
482
502

511
515
495
513

Candidates Who Pass

All Candidates
Currently
Formerly
Never

139,035 -95,268 69%

27,206 20%
16,561 12%

507
509
494
520

522
523
510
535

20,036 -9,326 47%

4,965 25%
5,745 29%

436
437
416
450

434
434
417
449

Candidates Who Fail

All Candidates
Currently
Formerly
Never
24

87%
91%
85%
74%

SAT
%
Pool

%
Pass
All Praxis II Test Takers
All Candidates
Currently
Formerly
Never

Math Verbal

%
Pass

89%
93%
85%
78%

ACT
%
Math English
Pool

111,520 -75,193 67%

21,459 19%
14,868 13%

19.7
19.9
19.2
19.2

21.5
21.7
21.0
20.8

99,650 -69,801 70%

18,272 18%
11,577 12%

20.1
20.2
19.7
20.0

22.0
22.1
21.7
22.1

11,870 -5,392 45%

3,187 27%
3,291 28%

16.4
16.4
16.3
16.3

16.7
16.6
16.9
16.5

Accreditation of teacher education programs has also been the target of

significant debate. The proportion of licensure candidates who report
attending institutions with NCATE-approved programs is higher for the ACT
pool than for the SAT pool. Overall, as displayed in Table 10, approximately
three fifths of those seeking licensure report having been trained at an
institution with an NCATE-approved program, a figure that is consistent with
other estimates of the size of the student population in NCATE-approved
institutions (Wise, 1997).
Passing rates for those who attend institutions with NCATE-approved
programs are higher than for those students who attend institutions not
accredited by NCATE. This holds for both SAT and ACT students. It is
especially noteworthy that the NCATE rates are higher since candidates at
NCATE institutions have somewhat lower SAT or ACT scores than do
candidates at non-NCATE schools. Thus, NCATE-accredited institutions
appear to increase the likelihood that candidates will meet state licensing
requirements.

Table 10: Mean SAT/ACT Scores by Passing Status and NCATE Accreditation (Praxis II)

160,075 -85,488 53%

74,587 47%

498
496
500

511
508
513

112,207 -77,048 69%

35,159 31%

19.7
19.6
19.9

21.5
21.5
21.5

Candidates Who Pass

All Candidates
NCATE
Non-NCATE

139,826 -77,771 56%

62,055 44%

507
503
512

522
517
528

89% 100,214 -91% 69,769 70%

87% 30,445 30%

20.1
19.9
20.4

22.0
22.0
22.2

20,249 -7,717 38%

12,532 62%

435
425
442

434
423
441

11,993 -7,279 61%

4,714 39%

16.3
16.2
16.5

16.6
16.6
16.8

Candidates Who Fail

All Candidates
NCATE
Non-NCATE

Math Verbal

%
Pass

ACT
%
Math English
Pool

%
Pass
All Praxis II Test Takers
All Candidates
NCATE
Non-NCATE

87%
91%
83%

SAT
%
Pool

25

Results by
Licensure
Area

Discussions of teacher academic ability most often treat teachers as a

homogeneous population. Given the varying kinds of academic skills required
of teachers in different licensure areas, we thought it would be useful to
examine college admissions scores by licensure area and see how these scores
compared with the scores of peers from the same subject matter disciplines.
For this analysis, we include only SAT scores.6
Figures 1 and 2 present SAT math and verbal data, respectively. In each figure,
we plotted the mean SAT scores for intended education majors and all college
bound seniors. We then plotted mean SAT scores for all college graduates.
These data appear as the three horizontal lines. We then plotted, for each
licensure area, the mean SAT score for those who passed the Praxis II tests.
The trends are fairly consistent for both math and verbal SAT, though teacher
candidate verbal performance is generally stronger than math performance.
There is a great disparity between those who qualify for licenses in specific
content areas and those who qualify for licenses in elementary, special, and
physical education. Elementary education candidates, by far the largest single
group of prospective teachers, have SAT scores that are clearly higher than
those of college bound seniors who announce their intention to be education
majors. Their verbal scores are slightly higher than college bound seniors,but
their math scores are somewhat lower. However, elementary education
candidates have much lower math and verbal scores than all college graduates.
For those who qualify for content licenses, we see a far different pattern. With
the exception of Art & Music, verbal SAT scores for teachers who succeed on
licensing tests in content areas are higher than for all college graduates, up to 35
points higher in the case of those passing licensing tests in English/Language
Arts. Math SATs for teacher candidates who pass licensing tests in
mathematics and science are substantially higher than for all college graduates
(55 and 26 points for math and science candidates, respectively). For virtually
all content areas, SAT scores for successful Praxis II candidates are significantly
higher than those for all college bound seniors.
Clearly, those qualifying for teacher licenses in specific content areas are a far
stronger group, academically, than is suggested by the intended education major
data. In fact, they are generally stronger than college bound seniors, and their
overall verbal skill as measured by the SAT exceeds mean performance of all
college graduates. Moreover, the math scores of teacher candidates who make
primary use of mathematics in their teaching far exceed the scores of the
average college graduate. However, claims of relatively low SAT scores
compared to their college peers still ring true for those qualifying for an
elementary, special education, or physical education license. And because
elementary teachers account for more than half of the teaching population, the
relatively low SAT scores of elementary teacher candidates pull down the
average SAT scores of the entire Praxis II population, thereby masking the
relatively high scores of those meeting licensure requirements in academic
content areas.

All relevant comparison ACT data were not available at the time of this report. We anticipate that such data will be
available and released at a later date.
6

26

Figure 1: Mean Math SAT Scores for Candidates Passing Praxis II by Licensing Area

Mean SAT Math Score

600

597

575

569

550

536
523

525

542

526

515
511
499

500
482

486
479

475
450

al
sic
y
Ph

n
on
tio
ati
c
ca
u
u
d
d
E
yE
ial
r
c
a
e
t
en
Sp
em
l
E

on
ati
c
u
Ed

t&
Ar

ic
us
M

ies
ud
t
lS
cia
o
S

sh
gli
En

ge
ua
g
n
La
n
g
rei
Fo

ce
ic en
S

cs
ati
m
e
ath
M

Licensure Area
Candidates Passing Praxis II
Intended Education Majors

College Bound Seniors

All College Graduates

27

Figure 2: Mean Verbal SAT Scores for Candidates Passing Praxis II by Licensing Area

Mean SAT Verbal Score

600
567

575

578
568

554

550

544
543

534

525
513

508

505

500

485

475

468

450

on
on
on
ati
ati
ati
c
c
c
u
u
u
d
Ed
Ed
lE
y
ial
r
c
is ca
e
ta
y
en
Sp
Ph
el m
E

t&
Ar

ic
us
M

cs
e
ies
d
ati
ag
u
u
t
m
g
S
e
n
l
ath
La
cia
o
n
M
g
S
rei
Fo

ce
ien
c
S

sh
gli
n
E

Licensure Area
Candidates Passing Praxis II
Intended Education Majors

28

College Bound Seniors

All College Graduates

We then explored the relationship among academic preparation, SAT scores,

and Praxis II passing rates. While 64% of those seeking elementary education
licensure majored in elementary education, 79-97% of those seeking licensure
in content domains had majors in either the content area or in the content
education area (e.g., science or science education). The proportion of those
with an undergraduate background in their content area varies dramatically by
licensure area.

Table 11: Mean SAT Scores and Passing Status by Undergraduate Major and Licensure
Area (Praxis II)
% Pass

% Pool

Math

Verbal

Art and Music

All Candidates
Art and Music Education Majors
Other Education Majors
Art and Music Majors
Non-Education Majors

91%
93%
82%
88%
68%

4,463
3,306
56
1,013
88

-74%
1%
23%
2%

515
515
513
515
540

534
532
523
540
564

Elementary Education
All Candidates
Elementary Education Majors
Other Education Majors
Non-Education Majors

87%
94%
75%
77%

43,399
27,926
2,365
13,108

-64%
5%
30%

499
486
493
528

513
498
506
549

English
All Candidates
English Education Majors
Other Education Majors
English Majors
Non-Education Majors

89%
92%
74%
91%
79%

7,471
2,688
197
3,577
1,009

36%
3%
48%
14%

526
508
512
535
543

578
556
562
591
596

Foreign Language
All Candidates
Foreign Language Majors
Other Education Majors

80%
83%
66%

1,765
1,440
325

-82%
18%

536
534
547

567
564
580

Mathematics
All Candidates
Mathematics Education Majors
Other Education Majors
Mathematics Majors
Non-Education Majors

76%
89%
50%
82%
56%

5,299
1,897
234
2,272
896

-36%
4%
43%
17%

597
580
581
604
623

544
524
532
548
577

29

Table 11 (continued)
% Pass

% Pool

Math

Verbal

Physical Education
All Candidates
Physical Education and Health Majors
Other Education Majors
Non-Education Majors

85%
88%
67%
64%

3,897
3,511
135
251

-90%
4%
6%

482
479
490
516

468
465
476
505

Science
All Candidates
Science Education Majors
Other Education Majors
Science Majors
Non-Education Majors

82%
92%
65%
84%
72%

4,470
268
193
3,749
260

-6%
4%
84%
6%

569
568
544
568
592

568
558
539
568
598

Social Studies Major

All Candidates
Social Studies Education Majors
Other Education Majors
Social Studies Majors
Non-Education Majors

75%
84%
65%
74%
63%

6,866
1,867
143
4,636
220

-27%
2%
68%
3%

523
509
508
528
554

554
534
542
560
593

Special Education
All Candidates
Special Education Majors
Other Education Majors
Non-Education Majors

91%
91%
89%
90%

9,004
4,948
2,233
1,823

-55%
25%
20%

486
482
481
503

508
502
501
532

Those with content area backgrounds have much higher passing rates than
candidates who do not have background in a content area. However, those
individuals with no content training have higher SAT scores overall than those
who have had content training. As with teacher education programs and
NCATE accreditation, there seems to be evidence that teacher education
institutions provide added value, in terms of licensing success, for candidates
who proceed through their programs. Given that most of the Praxis II tests
measure significant aspects of content, it is not surprising that pre-service
preparation in content is associated with greater success on licensure tests.

30

Raising and
Lowering
Standards

A critical debate in the teacher reform movement has been over raising or
lowering standards for entry into the profession. While some have argued that
raising standards will improve the overall academic ability of the teaching force,
others have countered that high standards will exacerbate supply problems,
particularly in less affluent school districts, and will further limit the
racial/ethnic diversity of the teaching force. In this section we look at how
raising or lowering scores by a relatively modest amountto the highest and
lowest passing scores established in any state in 1997would impact the
academic ability, supply, and diversity of the teaching force.
Each table in this section presents data at three passing score levels: the
hypothetical low passing score, the passing score in effect in a candidates state
at the time of the Praxis test administration, and the hypothetical high passing
score. Table 12 gives the data for Praxis I candidates by gender. The data
reveal that the passing rates vary drastically across standards.
At the low standard, more than 9 in 10 candidates would pass, but at the high
standard, passing rates drop below 50% for SAT takers and to 61 to 65% for
ACT takers. Thus, raising standards would clearly limit the supply of
prospective teachers gaining entry to teacher education programs. At all three
score levels, the impact essentially would be equal for males and females; it
does not appear that changing passing scores would alter the ratio of
approximately three females to every male in the prospective teaching
population.
Mean SAT/ACT scores rise commensurately with higher standards. Note,
however, that the difference in SAT/ACT scores is far greater between current
state passing standards and the high standards than it is between the state and
low standards. This suggests that the real payoff in increased academic ability
results from raising passing scores above their current levels, not from raising
them from a low standard up to the levels currently in place. However, this
increase comes at the cost of cutting the prospective teacher supply by almost
40% for all SAT takers, and approximately 30% for all ACT takers.

31

Table 12: Mean SAT/ACT Scores and Passing Status at Low, State, and High Passing
Standards, by Gender (Praxis I)
%
Pass

SAT
%
Pool

Math Verbal

%
Pass

ACT
%
Math English
Pool

Low Pass
All Candidates
Male
Female

92%
93%
91%

30,979
7,673 25%
23,306 75%

502
523
496

513
514
513

94%
95%
94%

52,053
14,160 27%
37,893 73%

20.1
20.9
19.8

21.2
20.4
21.5

State Pass
All Candidates
Male
Female

77%
79%
77%

26,182
6,537 25%
19,645 75%

514
535
507

525
527
524

88%
89%
87%

48,248
13,164 27%
35,084 73%

20.4
21.2
20.1

21.6
20.8
21.8

High Pass
All Candidates
Male
Female

47%
48%
47%

15,967
3,969 25%
11,998 75%

545
567
537

558
563
557

62%
65%
61%

34,011
9,670 28%
24,341 72%

21.4
22.1
21.1

22.8
22.0
23.2

Table 13 shows the results for modified Praxis I standards by race/ethnicity.

The results are similar to the gender data, in that SAT/ACT scores increase
substantially as passing standards are raised, and the largest SAT/ACT score
increases occur between the state and high passing standards. Also, passing
rates across all groups again drop substantially when scores are raised to the
high standard.

32

Table 13: Mean SAT/ACT Scores and Passing Status at Low, State, and High Passing
Standards by Race/Ethnicity (Praxis I)
%
Pass

SAT
%
Pool

Math Verbal

%
Pass

ACT
%
Math English
Pool

Low Pass
All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

91%
67%
92%
86%
84%
94%
87%

30,895
-2,415 8%
1,177 0%
517 1%
157 1%
26,111 85%
518 2%

502
442
526
477
474
507
508

513
456
500
489
477
519
527

95%
76%
89%
93%
92%
96%
93%

51,820
2,163
645
642
510
47,450
410

-4%
1%
1%
1%
92%
1%

20.1
16.9
21.2
18.9
17.6
20.3
20.2

21.2
17.8
20.3
19.7
18.5
21.4
21.1

State Pass
All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

77%
46%
76%
69%
64%
82%
71%

26,115 -1,650 6%
966 4%
418 2%
120 1%
22,537 86%
424 2%

514
456
542
490
490
517
526

525
474
517
506
497
529
547

88%
63%
77%
83%
81%
89%
84%

48,036
1,790
558
575
450
44,293
370

-4%
1%
1%
1%
92%
1%

20.4
17.2
21.7
19.2
17.8
20.5
20.6

21.5
18.4
21.1
20.1
18.9
21.7
21.6

High Pass
All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

47%
17%
45%
34%
28%
51%
46%

15,921 -612 4%
575 4%
202 1%
54 0%
14,200 89%
278 2%

545
496
567
532
519
546
554

558
513
552
551
548
560
580

62%
28%
54%
54%
48%
64%
61%

33,857
780
391
375
268
31,771
272

-2%
1%
1%
1%
94%
1%

21.4
18.4
22.8
20.3
18.8
21.5
21.8

22.8
20.2
22.5
21.4
20.5
22.9
23.1

Unlike for gender, passing rates vary substantially across racial/ethnic groups.
Thus, altering passing standards has implications for the racial/ethnic
composition of the candidate pool. For example, at all three passing score
levels, White candidates pass Praxis I at higher rates than all other groups, and
African American candidates pass at the lowest rates. At the low passing
standard, the vast majority of White candidates would pass Praxis I. The same
is not true for African American candidates, who would pass Praxis I at 67%
(SAT) and 76% (ACT).
Passing rate differences between racial/ethnic groups grow far larger as passing
scores are raised. This phenomenon is particularly apparent for the SAT group,
where more than half the White candidates would pass at the high standard, but
just 17% of the African American candidates would pass.

33

For both the SAT and ACT populations, the African American passing rate
drops by approximately 50 percentage points between the low and high passing
scores. The net effect of this drastic drop in passing rates is to cut the
percentage of those African American candidates in the teaching pool who take
Praxis I by about half. At the low passing score, 8% (SAT) and 4% (ACT) of
the passing population are African American, while at the high passing score,
these figures drop to 4% and 2%, respectively. It is worth reiterating, however,
that the prospective teaching force lacks racial/ethnic diversity even at the low
passing score, with 85% (SAT) and 92% (ACT) of the candidates in the low
passing pool being White. Further, because the majority of those who seek
licensure never even take the Praxis I tests, other factors must also contribute
to the homogeneity of the work force. Thus, entrance testing does not appear
to be the root cause for the lack of diversity of the workforce, but if done in
isolation, raising passing standards would clearly exacerbate an already troubling
trend.
Raising or lowering Praxis II passing scores would have much the same impact
as raising or lowering Praxis I passing scores. Table 14 contains the results by
gender. Regardless of the passing standard used, the overall composition of the
pool remains approximately three quarters female, even though females pass at
higher rates at the high passing standard.

Table 14: Mean SAT/ACT Scores and Passing Status at Low, State, and High Passing
Standards, by Gender (Praxis II)
%
Pass

SAT
%
Pool

Math Verbal

%
Pass

ACT
%
Math English
Pool

Low Pass
All Candidates
Male
Female

93%
93%
93%

149,037 -37,438 25%

111,599 75%

503
525
496

517
520
516

94%
93%
94%

105,564 -24,721 23%

80,843 77%

19.9
20.7
19.7

21.8
20.9
22.1

State Pass
All Candidates
Male
Female

87%
87%
88%

139,644 -34,697 25%

104,947 75%

507
529
500

522
525
521

89%
87%
90%

100,214 -23,274 23%

76,940 77%

20.1
20.8
19.9

22.0
21.1
22.3

High Pass
All Candidates
Male
Female

64%
59%
66%

102,546 -23,818 23%

78,728 77%

525
551
517

545
553
542

69%
63%
71%

77,408 -16,742 22%

60,666 78%

20.8
21.8
20.5

23.0
22.2
23.2

34

For both males and females, as the Praxis II passing scores increase, the
SAT/ACT scores increase as well. Of course, the higher the passing score, the
lower the supply of prospective teachers who meet that score. At the high
passing score, roughly two thirds of the Praxis II candidates would earn their
licenses, as opposed to well over 90% at the low passing score.
Breaking out the licensure test results by race/ethnicity provides much the
same picture as looking at the Praxis I data by race/ethnicity. As shown in
Table 15, nearly 9 in 10 candidates who meet the low passing score are White.
And more than 9 in 10 of those who pass, when the high passing standard is
applied, are White.
Once again we see the picture of a prospective teaching force that is not diverse
to begin with, and would be further restricted if minimum acceptable passing
standards were raised.

Table 15: Mean SAT/ACT Scores and Passing Status at Low, State, and High Passing
Standards, by Race/Ethnicity (Praxis II)

Low Pass
All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

%
Pass

SAT
%
Pool

Math Verbal

% Pass

ACT
%
Math English
Pool

93% 148,571
82%
9,469
84%
3,205
71%
3,819
89%
433
96% 129,021
85%
2,624

-6%
2%
3%
0%
87%
2%

503
434
543
479
486
508
507

517
453
522
492
493
522
530

94%
73%
90%
88%
87%
97%
84%

105,082
8,165
758
2,193
546
91,999
1,421

-8%
1%
2%
1%
88%
1%

19.9
17.1
22.9
18.8
18.5
20.2
19.4

21.8
18.6
22.2
20.0
20.2
22.2
21.0

State Pass
All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

87% 139,245
69%
7,984
75%
2,874
59%
3,134
80%
388
91% 122,534
76%
2,331

-6%
2%
2%
0%
88%
2%

507
441
547
486
491
511
515

522
463
529
503
499
525
541

89%
61%
82%
81%
81%
93%
74%

99,804
6,757
691
2,007
514
88,583
1,252

-7%
1%
2%
1%
89%
1%

20.1
17.4
23.0
18.8
18.6
20.3
19.7

22.0
19.0
22.7
20.1
20.4
22.3
21.5

High Pass
All Candidates
African American
Asian American/Asian
Hispanic
Native American
White
Other

64% 102,245
33%
3,837
59%
2,260
47%
2,506
54%
265
68% 91,520
60%
1,857

-4%
2%
2%
0%
90%
2%

525
466
563
498
512
527
532

545
497
550
517
525
547
563

69%
31%
70%
60%
51%
74%
55%

77,116
3,442
590
1,497
321
70,347
919

-4%
1%
2%
0%
91%
1%

20.8
18.3
23.8
19.9
19.9
20.9
20.8

23.0
20.5
23.4
21.3
22.0
23.1
22.9

35

At the current state passing scores, close to 9 in 10 of all candidates pass.

However, passing rates vary significantly across racial/ethnic groups, with
fewer than 60% of the Hispanic candidates from the SAT sample meeting their
state passing scores. In contrast, more than 80% of the Hispanic population
within the ACT sample meet their state passing scores. The reason for this
discrepancy is not readily apparent. We plan to explore these types of
racial/ethnic issues more closely in a future report by, among other things,
disaggregating results for groups within the various racial/ethnic categories.
Instituting high passing scores on licensure tests would have a significant
impact on the supply of teachers, as only about two thirds of all candidates
would pass. African American candidates would be the hardest hit by such an
increase, with only about one third passing at the high standard, compared to
more than two thirds of the White candidates. Across all racial/ethnic groups
and at all three passing standards, the licensure test passing rates are somewhat
higher than the entrance exam pass rates associated with Praxis I. So, for
institutions and states using the entire Praxis Series, the Praxis I test acts as a
barrier for proportionately more candidates than do the content-specific Praxis
II licensing tests. However, those who take entrance tests only comprise a
small proportion of those who eventually seek licensure, suggesting that other
forces also work to limit the diversity of the prospective teaching population.
One final way we looked at the potential impact of raising or lowering passing
scores was by examining candidatesundergraduate GPAs. The SAT/ACT
data from all the Praxis I and II gender and race/ethnicity tables provide strong
evidence that raising admissions and licensure testing standards will significantly
increase the academic caliber of the pool of teacher candidates. We chose to
look at GPA as well because we wanted to see the extent to which GPA would
substantiate our findings that used SAT/ACT scores as a proxy for academic
ability. To be sure, GPA data, especially when self-reported, must be
interpreted with caution, because they do not represent a standardized or highly
generalizable measure. Nonetheless, GPA does provide one indication of a
candidates academic performance as an undergraduate, thus making it a useful
complement to the SAT/ACT data that were collected on candidates when
they were college bound seniors.
Table 16 presents the findings with respect to GPA. As already discussed, the
vast majority of teaching candidates, even at the low passing score, have GPAs
above a 2.5, or Caverage. Since less than 1% of the candidates reported
GPAs below 2.0, these data are not reported. Most of the candidates at all
three levels fall between a 3.0 and 3.5 average. Note that passing rates rise with
GPA. This finding is not surprising, as one might expect that undergraduates
with stronger academic records would be more likely to pass licensing tests.
Between the low and state passing scores, passing rates are close enough that
raising the standard does not have much of an impact on the overall
composition of the pool. When the high standard is implemented, however,
the pool begins to assume a new shape, with respect to GPA. Fully one third
of the passing students at the high standard have a GPA of 3.5 or higher, and
less than 1 in 4 have a GPA below 3.0. This result occurs because more than 8
in 10 candidates with GPAs of at least 3.5 pass Praxis II even at the highest
standards, as opposed to fewer than half with GPAs below 3.0. These data
suggest that implementing the high passing standard would not pose a

36

particularly large barrier for those with GPAs above 3.5, but would serve to
deny licenses to at least 1 of every 3 people with lower GPAs. In contrast, at
the current state passing scores, the real differentiation appears to occur
between those candidates who have GPAs above 2.5 versus those with lower
GPAs.

Table 16: Mean SAT/ACT Scores and Passing Status at Low, State, and High Passing
Standards, by GPA (Praxis II)

Low Pass
All Candidates
3.5 - 4.0
3.0 - 3.49
2.5 - 2.99
2.0 - 2.49
State Pass
All Candidates
3.5 - 4.0
3.0 - 3.49
2.5 - 2.99
2.0 - 2.49
High Pass
All Candidates
3.5 - 4.0
3.0 - 3.49
2.5 - 2.99
2.0 - 2.49

%
Pass

93% 148,646
98% 39,595
94% 62,708
90% 41,427
80%
4,845

SAT
%
Pool

Math Verbal

%
Pass

ACT
%
Math English
Pool

-27%
42%
28%
3%

503
532
503
479
475

517
554
516
486
479

94%
98%
96%
91%
79%

105,114
28,001
42,656
30,380
3,961

-27%
41%
29%
4%

19.9
21.7
19.9
18.6
17.9

21.8
24.1
21.8
20.1
19.4

139,217
38,439
59,350
37,338
4,029

-28%
43%
27%
3%

507
534
506
484
483

522
556
520
492
489

89%
96%
92%
84%
70%

99,801
27,438
40,765
28,020
3,487

-27%
41%
28%
3%

20.1
21.8
20.0
18.8
18.1

22.0
24.2
22.0
20.3
19.7

64% 102,580
83% 33,534
66% 44,029
49% 22,555
40%
2,422

-33%
43%
22%
2%

525
544
522
505
503

545
569
541
519
515

69%
86%
72%
56%
39%

77,101
24,439
32,007
18,645
1,963

-32%
42%
24%
3%

20.8
22.1
20.6
19.4
19.2

23.0
24.7
22.7
21.4
21.2

87%
95%
89%
81%
67%

As one would expect, SAT/ACT scores rise with both passing standards and
GPA. In fact, for the top GPA group at the high passing standard, the SAT
math scores are equivalent to those of all college graduates, and the SAT verbal
scores are significantly higher than those for college graduates.
The low, state, and high passing score data by gender, race/ethnicity, and GPA
confirm two important points. First, the higher states choose to set passing
scores on entrance and licensure tests, the better the academic quality of their
pool will be. Second, this academic improvement clearly comes at the cost of
limiting both the supply and diversity of the potential teaching force.

37

CONCLUSIONS
Taken together, our data highlight the complexity of the issues facing
policymakers as they debate critical issues of teaching and learning. Based on
this study, we make the following assertions about teaching candidates and
teacher testing.
First, the teacher population hardly can be considered an academically
homogeneous group. Sweeping claims about the academic skills of all teachers
do not take into account the profound differences between groups of
individuals seeking different kinds of licenses. Policy implications, in turn, are
quite different for different segments of the prospective teaching force. For
elementary teachers, the issue seems to be one of improving quality, to the
extent that SAT/ACT scores are an appropriate measure of quality. For
content specialists, the issue appears to be one of increasing quantity.
Second, candidates applying for a teaching license, specifically those who
pursue licensure in academic content areas, have reasonably strong academic
skills skills that are substantially higher than many have suggested. These
individuals have academic skills, at least as measured by college admissions
tests, that generally are as strong as, and in several cases stronger than, those of
their college graduate peers.
Third, the relatively poor standing of candidates in the elementary, special
education, and physical education teacher pools supports reform efforts that
argue that all teachers ought to have a firm grounding in at least one academic
content area. Those with strong academic skills do not seem to be attracted to
elementary education from the outset. And, since elementary education
teachers make up, by far, the largest group of licensed teachers, this issue
cannot be minimized. In fact, our data clarify that general claims about
teachers are dominated by the overwhelming proportion of elementary teachers
in any representative sampling.
Fourth, admissions and licensure testing clearly raise the academic ability of the
population by denying access to those with lower test scores. Without
exception, we found that those who passed Praxis I and II tests had higher
average SAT/ACT scores than those who failed, and that the higher the Praxis
passing score was set, the higher the SAT/ACT scores were of the passing
population. Teacher testing thus appears to be working in the sense that it
elevates the academic ability of the pool of teachers.
Fifth, the effect of testing on the diversity of the teaching force is not
promising. Worse, the lack of diversity cannot be simply ascribed to testing
policies. The proportion of minority individuals being attracted to teacher
education is far smaller than the proportion of minority students in U.S.
classrooms. Disparate passing rates by race/ethnicity exacerbate this mismatch
between the teacher and student populations. Licensure testing takes a
predominantly White population of potential teachers and creates an even more
homogeneous group. Our data suggest that, without radical changes in the
recruitment and adequate training of talented minorities, this trend will not
change any time soon.

38

Sixth, the raising of licensing standards has profound effects on the

characteristics of the pool of prospective teachers. Raising standards raises the
academic profile of those who meet passing requirements by reducing the pool
of candidates and selectively removing individuals with less academic skill.
Raising standards also limits the racial/ethnic diversity of the pool of
prospective teachers who meet passing requirements. We do not argue that the
impact on diversity implies that passing standards should not be raised; indeed,
the data in this report provide convincing evidence that high standards lead to a
teacher population with even higher academic ability. We do feel, however,
that bleak passing rates and disparities among racial/ethnic groups suggest that
any moves to adopt higher standards must be accompanied by aggressive
efforts to support and enhance all candidatesknowledge and abilities so they
stand a better chance of meeting those high standards.
Seventh, findings based on undergraduate GPA data mirror the SAT/ACT
results. Students who are more successful in the classroom also are more
successful on licensing tests and have higher college admissions test scores.
Raising standards increases the likelihood that the pool of teachers will be made
up of individuals with strong academic records.
Eighth, teacher education programs, and NCATE accredited institutions, in
particular, are relatively successful in helping students meet licensure testing
requirements. Passing rates for students attending these institutions are higher
than for students from other institutions, even when students from other
institutions have higher mean college admissions scores.
Ninth, while there are no ready answers to the problems facing education, and
teacher preparation in particular, it is clear some frequently cited data do not
bring relevant information to the debate. For example, using intended
education major as a proxy for those who enter the teaching field is not helpful.
There is little relationship between those who actually meet licensure
requirements and those who think they might major in education as they
complete high school. It is also clear, in light of SAT scores, that assessments
of the teacher pool should account for the relative proportion of females in the
sample, as women have traditionally scored less well than males on these tests,
even when other measures suggest that they have at least equal academic ability.
If high standards are the wave of the future, and passing scores on Praxis and
other teacher tests are to rise, effective ways to increase both the overall supply
of teachers and the relative percentages of minority teachers must be found,
particularly in specific content areas. Many reforms have been suggested in this
vein, such as higher teacher salaries, targeted recruitment and training as early as
middle school, rigorous but flexible alternate route programs to lure mid-career
professionals into teaching, and supportive induction programs to lower
attrition rates of novice teachers.
While all these sound promising in theory, it remains to be seen whether they
and similar measures will be effective on a large scale. Until they are,
policymakers must walk a tightrope with respect to teacher testing. Our data
suggest that the mere act of raising passing scores will not be a silver bullet
solution for improving teacher quality. Rather, these data suggest that though
testing with higher standards holds great promise for ensuring that teachers are
academically able, if not used judiciously, such testing can also exacerbate
already daunting problems with the supply and diversity of potential

39

teachers. Those who both major in content areas and complete teacher
education programs appear to be a relatively strong and academically able
group. The challenge now is to create incentives and mechanisms that
increasingly promote this pathway and to avoid pathways that neither attract
nor produce a pool of racially/ethnically diverse teacher candidates with strong
academic skill.

40

REFERENCES
ACT. (1997). The high school profile report: Normative data. Iowa City, IA: Author.
American Association of Colleges for Teacher Education (in press). Teacher education pipeline IV: Schools, colleges, and
departments of education enrollments by race, ethnicity, and gender. Washington, DC: Author.
Ballou, D. (1996). Do public schools hire the best applicants? Quarterly Journal of Economics, 111, 97-134.
Ballou, D., & Podgursky, M. (1998). The case against teacher certification. Public Interest, 132, 17-29.
Boe, E. E., & Gilford, D. M. (1992). Summary of conference proceedings. In E. E. Boe & D. M. Gilford (Eds.),
Teacher supply, demand, and quality: Policy issues, models, and data bases (pp. 21-60). Washington, DC: National
Academy Press.
Carnegie Forum on Education and the Economy. (1986). A nation prepared: Teachers for the 21st century. Washington,
DC: Author.
Coleman, J. S., Campbell, E. Q., Hobson, C. J., McPartland, J., Mood, A. M., Weinfeld, F. D., & York, R. L. (1966).
Equality of educational opportunity. Washington, DC: U.S. Office of Education.
Darling-Hammond, L. (1998). Teachers and teaching: Testing policy hypotheses from a National Commission
report. Educational Researcher, 27(1), 5-15.
Darling-Hammond, L., & Cobb, V. L. (1996). The changing context of teacher education. In F. B. Murray (Ed.), The
teacher educator's handbook: Building a knowledge base for the preparation of teachers (pp. 14-62). San Francisco:
Jossey-Bass.
Darling-Hammond, L., Dilworth, M. E., & Bullmaster, M. (1996). Educators of color (commissioned paper).
Washington, DC: Office of Educational Research and Improvement (OERI), US Department of
Education.
Darling-Hammond, L., & Wise, A. E. (1983). Teaching standards, or standardized teaching? Educational Leadership,
41(2), 66-69.
Ehrenberg, R. G., & Brewer, D. J. (1995). Did teachers' verbal ability and race matter in the 1960s? Coleman
revisited. Economics of Education Review, 14(1), 1-21.
Ekstrom, R. B., & Goertz, M. E. (1985, March). The teacher supply pipeline: The view from four states . Paper
presented at the annual meeting of the American Educational Research Association, Chicago, IL. (ERIC
Document Reproduction Service No. ED 266 118)
Feistritzer, C. E. (1996). Profile of teachers in the U.S. Washington, DC: National Center for Education Information.
Ferguson, R. F. (1991). Paying for public education: New evidence on how and why money matters. Harvard Journal
on Legislation, 28, 465-498.
Ferguson, R. F. (1998). Can schools narrow the Black-White test score gap? In C. Jencks & M. Phillips (Eds.), The
Black-White test score gap (pp. 318-374). Washington, DC: The Brookings Institution.
Fox, J. (1998, October 12). Teacher incubation time cut back in California. Education Daily, pp. 1-2.
Galambos, E. C. (1985). Teacher preparation: The anatomy of a college degree. Atlanta, GA: Southern Regional Education
Board.
Gardner, W. E., Scannell, D., & Wisniewski, R. (1996). The curious case of NCATE redesign. Phi Delta Kappan, 77,
622-629.
Goertz, M. E., Ekstrom, R. B., & Coley, R. J. (1984). The impact of state policy on entrance into the teaching profession (Final
Report under National Institute of Education Grant No. G83-0073). Princeton, NJ: Educational Testing
Service.
Goertz, M. E., & Pitcher, B. (1985). The impact of NTE use by states on teacher selection (Research Report No. RR-85-1).
Princeton, NJ: Educational Testing Service.
Greenwald, R., Hedges, L. V., & Laine, R. D. (1996). The effect of school resources on student achievement. Review
of Educational Research, 66, 361-396.
Grossman, P. L., Wilson, S. M., & Shulman, L. S. (1989). Teachers of substance: Subject matter knowledge for
teaching. In M. C. Reynolds (Ed.), Knowledge base for the beginning teacher (pp. 23-36). New York: Pergamon
Press.
Haney, W., Madaus, G., & Kreitzer, A. (1987). Charms talismanic: Testing teachers for the improvement of
American education. In E. Z. Rothkopf (Ed.), Review of research in education (Vol. 14, pp. 169-238).
Washington, DC: American Educational Research Association.
Hanushek, E. A., & Pace, R. R. (1994). Understanding entry into the teaching profession. In R. G. Ehrenberg (Ed.),
Choices and consequences: Contemporary policy issues in education (pp. 12-28). Ithaca, NY: ILR Press.
Hanushek, E. A., & Pace, R. R. (1995). Who chooses to teach (and why)? Economics of Education Review, 14, 101-117.

41

Hedges, L. V., & Greenwald, R. (1996). Have times changed? The relation between school resources and student
performance. In G. Burtless (Ed.), Does money matter? The effect of school resources on student achievement and adult
success (pp. 74-92). Washington, DC: Brookings Institution Press.
Hirsch, E., Koppich, J. E., & Knapp, M. S. (1998). What states are doing to improve the quality of teaching: A brief review of
current patterns and trends (A CTP working paper ). Seattle: University of Washington, Center for the Study of
Teaching and Policy.
Howey, K. (1996). Designing coherent and effective teacher education programs. In J. Sikula, T. J. Buttery, & E.
Guyton (Eds.), Handbook of research on teacher education (2nd ed., pp. 143-170). New York: Macmillan.
Ingersoll, R. M. (1996). National assessments of teacher quality (NCES-WP-96-24). Washington, DC: U.S. Department of
Education, Office of Educational Research and Improvement, National Center for Education Statistics.
Kerr, D. H. (1983). Teaching competence and teacher education in the United States. In L. S. Shulman & G. Sykes
(Eds.), Handbook of teaching and policy (pp. 126-149). New York: Longman.
Koerner, J. D. (1963). The miseducation of American teachers. Boston: Houghton Mifflin.
Lanier, J. W. (1986). Research on teacher education. In M. C. Wittrock (Ed.), Handbook of research on teaching (3rd ed.,
pp. 527-569). New York: Macmillan.
Lee, J. B. (1984). Tomorrow's teachers . Washington, DC: U.S. Department of Education. (ERIC Document
Reproduction Service No. ED 346 082)
Murnane, R. J. (1991). The case for performance-based licensing. Phi Delta Kappan, 73, 137-142.
Murnane, R. J., Singer, J. D., Willett, J. B., Kemple, J. J., & Olsen, R. J. (1991). Who will teach? Policies that matter.
Cambridge, MA: Harvard University Press.
National Association of State Boards of Education. (1998). The numbers game: Ensuring quantity and quality in the teaching
workforce. Alexandria, VA: Author.
National Association of State Directors of Teacher Education and Certification. (1998). Manual on certification and
preparation of educational personnel in the United States and Canada. (3rd ed.). Dubuque, IA: Kendall/Hunt.
National Center for Education Statistics. (1995). America's Teachers Ten Years After "A Nation At Risk" (NCES 95-766).
Washington, DC: U.S. Department of Education, Office of Educational Research and Improvement.
National Center for Education Statistics. (1996). Digest of education statistics 1996 (NCES 96-133 ). Washington, DC:
U.S. Department of Education, Office of Educational Research and Improvement.
National Center for Education Statistics. (1997a). America's teachers: Profile of a profession, 1993-94 (NCES 97-460 ).
Washington, DC: U.S. Department of Education, Office of Educational Research and Improvement.
National Center for Education Statistics. (1997b). Characteristics of stayers, movers, and leavers: Results from the teacher
followup survey: 1994-95 (NCES 97-450). Washington, DC: U.S. Department of Education, Office of
Educational Research and Improvement.
National Center for Education Statistics. (1998a). Digest of education statistics 1997. (NCES 98-015). Washington, DC:
U.S. Department of Education, Office of Educational Research and Improvement.
National Center for Education Statistics. (1998b). Projections of education statistics to 2008. (NCES 98-016). Washington,
DC: U.S. Department of Education, Office of Educational Research and Improvement.
National Center for Education Statistics. (1999). Teacher quality: A report on teacher preparation and qualifications of public
school teachers (NCES 1999-080). Washington, DC: U.S. Department of Education, Office of Educational
Research and Improvement.
National Commission on Excellence in Education. (1983). A nation at risk: The imperative for educational reform.
Washington, DC: Author.
National Commission on Teaching & Americas Future. (1996). What matters most: Teaching for Americas future. New
York: Author.
National Commission on Teaching and America's Future. (1997). Doing what matters most: Investing in quality teaching.
New York: Author.
Nicklin, J. L. (1992, May 6). Teacher-education programs debate the need for accrediting agency's stamp of approval.
The Chronicle of Higher Education, 38, A19, A22.
Reynolds, A. (1992). What is competent beginning teaching? A review of the literature. Review of Educational Research,
62, 1-35.
Riley, R. W. (1998). The challenge for America: A high quality teacher in every classroom. Annual Back to School Address,
National Press Club, Washington, DC, September 15, 1998 [On-line]. Available:
http:/www.ed.gov/Speeches/980915.html.
Roth, R. A. (1996). Standards for certification, licensure, and accreditation. In J. Sikula, T. J. Buttery, & E. Guyton
(Eds.), Handbook of research on teacher education (2nd ed., pp. 242-278). New York: Macmillan.
Schlechty, P. C., & Vance, V. S. (1983). Recruitment, selection, and retention: The shape of the teaching force. The
Elementary School Journal, 83, 469-487.

42

Shanker, A. (1996). Quality assurance: What must be done to strengthen the teaching profession. Phi Delta Kappan,
78, 220-224.
Shulman, L. S. (1987). Knowledge and teaching: Foundations of the new reform. Harvard Educational Review, 57, 1-22.
Smith, G. P. (1987). The impact of competency tests on teacher education: Ethical and legal issues in selecting and
certifying teachers. In M. Haberman & J. M. Backus (Eds.), Advances in teacher education (Vol. 3, pp. 218-249).
Norwood, NJ: Ablex.
Sowell, T. (1993). Inside American education: The decline, the deception, the dogmas. New York: The Free Press.
Strauss, R. P., & Sawyer, E. A. (1986). Some new evidence on teacher and student competencies. Economics of
Education Review, 5, 41-48.
Sykes, G. (1983). Public policy and the problem of teacher quality: The need for screens and magnets. In L. S.
Shulman & G. Sykes (Eds.), Handbook of teaching and policy (pp. 97-125). New York: Longman.
The College Board. (1997). College bound seniors: A profile of SAT program test takers . New York: Author.
The Holmes Group. (1986). Tomorrow's teachers. East Lansing, MI: Author.
Toch, T. (1991). In the name of excellence: The struggle to reform the nation's schools, why it's failing, and what should be done. New
York: Oxford University Press.
U.S. Department of Education. (1998). Promising practices: New ways to improve teacher quality. Washington, DC: Author.
Weaver, W. T. (1983). America's teacher quality problem: Alternatives for reform. New York: Praeger.
Wise, A. E. (1997). NCATE: Making a difference in teacher preparation. Principal, 77(1), 12-13.
Wise, A. E., & Leibbrand, J. (1996). Profession-based accreditation: A foundation for high-quality teaching. Phi Delta
Kappan, 78, 202-206.
Zimpher, N. L., & Yessayan, S. (1987). Recruitment and selection of minority populations into teaching.
Metropolitan Education, 5, 57-71.

43

APPENDIX A
The following table lists all the states that had established passing scores on one
or more of the Praxis I and/or Praxis II tests in 1997. Praxis administration
volumes vary widely across states, as some states may elect to use only a small
subset of tests within the series, whereas others may use a large number of
different Praxis test titles

.
State
Arizona
Arkansas
California
Connecticut
Delaware
District of Columbia
Florida
Georgia
Hawaii
Indiana
Kansas
Kentucky
Louisiana
Maine
Maryland
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Jersey
New Mexico
New York
North Carolina
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
Tennessee
Virginia
West Virginia
Wisconsin

Praxis I
X
X
X
X
X
X
X
X

X
X
X
X
X
X

X
X
X
X

X
X
X
X

Praxis II
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X

X
X
X
X
X
X
X
X
X
X
X
X
X

44

ETS and the Teaching and Learning Division

are committed to research efforts to
improve the quality of these assessments
and contribute to understanding
teaching and teacher assessment
more broadly.

The Teaching and Learning

Research Report Series
is published by the
Teaching and Learning Division
of Educational Testing Service.
The Division works on a wide range
of teacher assessments, including licensure,
advanced certification, and formative assessments
for teacher professional development.

52311-16293 U59M4.4 Printed in U.S.A.

I.N. 225033